Re: Bag of Binary Words VS RTABMap

Posted by leo152 on
URL: http://official-rtab-map-forum.206.s1.nabble.com/Bag-of-Binary-Words-VS-RTABMap-tp2335p2357.html

Hi!
   I really appreciate for your response. There is still some problems I cannot figure out.
   
  1. For incrementalFlann, if you use this method, the vocabulary should be reconstructed after each image. However, it appears that if  Kp/IncrementalFlann is set false, the vocabulary is updated after each image.
  2. We get the same results with / without  Kp/IncrementalFlann, so why do we need this time-consuming module?
  3. I tried the parameters you metioned, and the input is like this :
./rtabmap-console -Rtabmap/StatisticLogged true -Rtabmap/StatisticLoggedHeaders false -Rtabmap/PublishStats false -Mem/RehearsalIdUpdatedToNewOne true -Mem/BadSignaturesIgnored true -Rtabmap/LoopRatio 0.0  -Mem/STMSize 30 -Mem/RehearsalSimilarity 0.2 -Mem/RecentWmRatio 0.2 -SURF/Extended false -Kp/WordsPerImage 400 -Kp/BadSignRatio 0.25 -SURF/HessianThreshold 1000 -Kp/TfIdfLikelihoodUsed true -Bayes/FullPredictionUpdate false -Rtabmap/TimeThr 1400 -Kp/DetectorStrategy 2 -Kp/NNStrategy 3 -Kp/NndrRatio 0.9 ~/Documents/data/Images_CityCentre/_joined/
    I used orb features since it's fast to compute, However, the computation time reached 1s per image at the 400th iteration. Other NNStrategy performs poorly for binary features.
    When I changed to surf features, the runtime is pretty steady at around 300ms per image.